gnash Bug with apport Enabled Results in Very High Run Queue
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
apport (Ubuntu) |
New
|
Undecided
|
Unassigned |
Bug Description
Binary package hint: apport
Ubuntu 7.10. apport 0.98.
If bug #214537 is re-created with apport enabled, i.e. "/etc/init.d/apport start", then once the disc activity starts, presumably because the machine is getting low on RAM, it never stops. The machine gets very unresponsive, mouse movement stops, etc. It doesn't improve despite leaving it for over ten minutes. Using the Magic Alt-SysRq key I can kill off enough to get to a console, there w(1) reports system load averages (e.g. run queue) of over 25 processes for the previous 1, 5, and 15 minutes. This is on a single-user machine doing very little.
Turning apport off and then re-creating the bug gives the behaviour described in the bug report, i.e. a few seconds of disc activity before firefox and gnash are killed by the kernel's OOM killer.
Conclusion, apport trying to handle the situation makes it far, far, worse. I doubt the machine would ever have returned to normal. It seems apport doesn't detect its work isn't going well and ends up creating lots of run queue processes.
This has been marked as a dupe of bug #207629. I spotted that one before opening this one, but it was with 8.04 pre-release whereas I'm on 7.10. Given the comment on bug #207629 that the plan is to disable apport by default just prior to release, why doesn't that apply to 7.10? It seems 7.10 users will be left by default with a system that dies every so often unless they investigate far enough themselves to disable apport. Could not apport be disabled for them in an update?